基于视觉词典的多目标截面投影图像特征分割

Multi-objective Cross-sectional Projection Image Feature Segmentation Based on Visual Dictionary

由于传统截面投影图像特征分割方法存在噪声免疫力低、适用广泛性窄和分割表现力差等诸多问题,因此提出一种基于视觉词典的多目标截面投影图像特征分割。使用k-means法完成聚类并利用贪婪法融合结果形成视觉词典,通过构建三维直方图,将像素点投影到主斜线上获得基本极大类间方差准则,实现像素与像素间的分类,同时为了增强噪声免疫力将阈值转化成可以变化的数值,通过峰值信噪比的最大比率对阈值合理取值,完成多目标截面投影图像特征分割。通过仿真可知,合理的阈值能够有效降低图像像素的错分,使图像特征分割更加精准,且不计附带的噪音种类,与其它方法对比发现,所提方法不仅噪声免疫力强且适用范围广,分割效果极佳。

Traditionally, the cross-sectional projection image feature segmentation method has some problems such as low noise immunity, low universality and poor segmentation performance. Therefore, this article presented a method to segment multi-objective cross-sectional projection image based on visual dictionary. Firstly, the k-means method was used to complete the clustering and the greedy algorithm was used to integrate with the results to form a visual dictionary. By constructing a three-dimensional histogram, the pixel points were projected on the main oblique line to obtain the basic criterion of maximal inter-class variance. Thus, the classification between pixels was achieved. In order to enhance the noise immunity, the threshold was converted into a variable value. Finally, the max ratio of peak signal-to-noise ratio was used to reasonably calculate the threshold value. Thus, the segmentation for multi-target cross-sectional projection image feature was completed. Through simulation results, we can see that the reasonable threshold can effectively reduce the misclassification of image pixels, so that the image feature segmentation will be more accurate, and the incidental noise types can be ignored. Compared with other methods, the proposed method has strong noise immunity and wide applicability. Its segmentation effect is excellent.